GenAI browser assistants risk user privacy, study finds

GenAI browser assistants may pose privacy risks, collecting and sharing sensitive data without user knowledge, UC Davis study reveals.

Why it matters

  • A UC Davis study shows GenAI browser assistants can collect and share sensitive user data without consent.

By the numbers

  • 9 GenAI browser assistants were studied, including Monica, Sider, ChatGPT for Google, Merlin, MaxAI, Perplexity, HARPA.AI, TinaMind and Copilot.
  • One extension, Merlin, was found to collect form inputs, including social security numbers.

The big picture

  • The study highlights the need for stronger safeguards and transparency to protect user privacy.
  • Some assistants shared data with third-party servers, like Google Analytics, raising concerns about targeted ads and profiling.

What they're saying

  • Users express concerns about AI integration and data privacy, with some drawing parallels to Windows Recall.
  • Some commenters note that data collection is a longstanding internet business model.

Caveats

  • The study was conducted on a specific set of GenAI assistants and may not represent all available options.

What’s next

  • Researchers suggest that addressing these risks will require efforts across the GenAI ecosystem and user awareness.